Optimization method of surface mounting process of in-line placement machine based on hierarchical heuristics

Abstract

The invention relates to the field of PCB assembly process optimization based on hierarchical heuristics. The invention solves the problems of low assembly efficiency caused by strong application scenario limitation, slow optimal path solution speed, and incomplete consideration of factors affecting assembly efficiency in the existing PCB mounting optimization technology of the in-line mounting machine. The invention first allocates each component type to the corresponding slot, considers the status information of each slot in the process of allocating the corresponding slot, and allocates the slot under the constraint condition of allocating the optimal slot to the mounting head at the head end of the in-line mounting head mechanism; secondly, allocates the corresponding component type to each mounting head, and finally obtains the optimal mounting path under the current allocation result according to the constraint condition of the component type allocation result of each mounting head on the in-line mounting head mechanism in each picking cycle according to the nearest neighbor cluster search strategy. The invention is mainly used for the optimization of PCB assembly process.

Description

Technical Field

[0001] The present invention relates to the field of PCB assembly process optimization. Background Art

[0002] Nowadays, with the widespread application of electronic products in modern life, the electronic manufacturing industry has developed rapidly. The manufacturing process of electronic products consists of a series of complex production processes, and PCB assembly is one of the most important and time-consuming processes. PCB assembly refers to a series of process flows for installing components in pre-specified positions on circuit boards, which is usually completed by a placement machine. A placement machine is a complex computer-controlled PCB assembly device that integrates multiple technologies including mechanics, electrical, optical, and operations research.

[0003] The placement machine is divided into many types according to the assembly process and mechanical structure. The schematic diagram of the in-line single cantilever placement machine is as follows: Figure 1 As shown. The main difference between an in-line placement machine and other types of placement machines is that multiple placement heads are arranged linearly to simultaneously complete component picking actions and improve assembly efficiency. This type of placement machine usually includes a fixed PCB platform, two fixed feeder bases, and a movable cantilever with a placement head. The feeders with components are mounted on the front and rear feeder bases. The cantilever can reciprocate between the PCB and the feeder, wherein each feeder on the feeder base is used to store one type of component. The placement head equipped with a Z-axis suction rod and a suction nozzle picks up components from the feeder through the switch of the vacuum valve and places the components at the predetermined position of the PCB. The flying camera equipped on the placement head is used to detect the chip picking offset. For some larger chips, the placement head needs to move to the fixed camera for detection. In addition, different types of nozzles are stored in the ANC (Auto Nozzle Changer), and the placement head completes the nozzle replacement action in the ANC to meet the assembly requirements of the placement head for different component shapes.

[0004] The process flow of surface mounting process in the prior art is as follows: Figure 2As shown. The specific process of the placement machine to complete the assembly of the circuit board is: the operator first needs to determine the type of assembly components according to the BOM (Bill of Material) table, install the components in the feeder, and insert the feeder into the base slot through manual designation or automatic optimization results. The circuit board is transferred to the assembly position by the conveyor belt, and then the placement head checks the consistency of the component type to be mounted and the nozzle type, and goes to the ANC to replace the nozzle if necessary. In a pick-up cycle, the placement head goes to the feeder base to pick up the component, detects the component picking offset through the visual system, and mounts it to the specified position on the circuit board. In the PCB assembly process, the pick-up cycle is the basic unit, which refers to the process of a group of placement heads completing component picking, identification, and placement (waste). The present invention does not discuss the impact of the waste process on the assembly efficiency, that is, it is assumed that there is no situation of picking failure and identification failure. For different types of circuit boards, the types of components and placement point positions to be mounted are also different. Different feeder installation positions and component placement sequences will affect the assembly efficiency of the placement machine.

[0005] The relationship logic diagram of the surface mounting process optimization problem is as follows Figure 3 As shown in the figure, existing research decomposes the surface mounting optimization task into the feeder slot allocation problem and the placement head task allocation problem. The former solves the feeder installation slot problem, while the latter determines the assembly order of the placement head. The sub-problems of the PCB assembly process are closely coupled. The feeder installation slot affects the component allocation process, which in turn affects the maximum number of simultaneous picks. The nozzle type of each placement head picking up components is directly related to the number of nozzle changes. The material picking slot and placement order of the surface mounting process jointly determine the overall movement distance of the cantilever. In order to ensure the consistency of nozzle type, component type, placement point and feeder slot, the placement head may generate redundant picking and nozzle replacement movements during the PCB assembly process.

[0006] The goal of the surface mount process is to minimize the assembly time, which depends on the distance the cantilever moves, the number of simultaneous picks, and the number of nozzle changes, which are the sub-goals of surface mount task optimization. The coupling of sub-goals is reflected in the fact that increasing the number of simultaneous picks may lead to additional nozzle changes, and reducing the number of nozzle changes may lead to an increase in the pick path and the number of picks. The optimization algorithm must make a trade-off between the pick efficiency and the nozzle change efficiency. The cantilever's moving distance is directly related to the path planning of the assembly process. At the same time, increasing the number of simultaneous picks can also help reduce the moving distance. Excessive nozzle changes require the cantilever to frequently go back and forth between the ANC and increase the moving path length. For in-line placement head placement machines, the picking efficiency and the number of nozzle changes are important factors affecting the overall placement efficiency. Optimizing these two sub-goals will also help reduce the cantilever's moving distance. In the actual optimization process, the surface mount process optimization method is usually regarded as a multi-objective optimization problem, that is, multiple optimization sub-goals will affect the overall placement efficiency, and the performance indicators between them conflict with each other. Specifically, when processing simultaneous picking, in order to ensure that multiple placement heads complete the picking action from the predetermined slots at the same time, it is necessary to adjust the nozzle type according to the actual situation to meet the picking requirements; in order to reduce the number of picking cycles, the placement head needs to ensure that the cantilever picks up as many components as possible in each picking cycle, which will indirectly lead to an increase in the moving path and a reduction in picking.

[0007] The placement process optimization algorithm must not only deal with various types of components and placement points corresponding to different PCB data, but also consider the constraints in the actual application process. The constraints of the surface mount optimization problem can be divided into four parts: work integrity constraints, mechanical limitation constraints, tool constraints and human constraints. Work integrity constraints are the basic requirements of surface mounting tasks. Each component must be accurately assembled on the corresponding PCB pad; mechanical limitation constraints refer to the constraints brought about by the structural characteristics of the placement machine, such as each placement head cannot reach the feeder slot, and the installation slot position interference caused by the different widths of the feeder; tool constraints include quantity constraints and consistency constraints. Quantity constraints refer to the number of nozzles used to grab components and the number of feeders that provide components. Consistency constraints refer to the need to match the nozzle type with the component to ensure the smooth progress of the picking process. Inappropriate nozzle type will lead to grasping failure, high scrap rate, etc., which will further reduce assembly efficiency; in addition, artificial constraints refer to additional restrictions added by on-site operators based on actual conditions, such as setting some feeder slots, disabling placement heads or slots, changing the number of available feeders, specifying the type of nozzles used by placement heads, etc. Artificial constraints are usually appropriate adjustments made based on the on-site machine conditions and are necessary to ensure the smooth progress of the production process.

[0008] The main problems in the existing in-line placement machine optimization technology for PCB placement process are as follows:

[0009] (1) The optimization solution for placement efficiency is not ideal. Specifically, during the PCB placement process of the in-line placement machine, the number of simultaneous pick-ups by the placement head group, the number of nozzle changes of the placement head, and the length of the cantilever's moving path will all affect the assembly efficiency. Relevant research only focuses on one or two of these indicators, resulting in low placement efficiency.

[0010] (2) There are limitations in actual application scenarios. When solving the feeder allocation problem, related studies have found that when some slots on the feeder base are vacant and there are pre-existing components in the remaining slots, all slots must be emptied before use, and the component types must be reallocated in the feeders on all slots. This allocation method has limitations and is not suitable for small-batch fixed feeder configuration scenarios, or scenarios where data needs to be frequently adjusted and re-optimized, which will increase labor costs.

[0011] (3) The problem of slow optimal solution speed in the process of optimizing the path. The PCB mounting process optimization methods represented by mathematical programming and evolutionary algorithms have the problems of slow optimal path solution speed and poor convergence.

[0012] To sum up, the present invention aims to conduct in-depth research and further enrich relevant research details, optimize the surface mounting process of PCB by the in-line placement machine, improve the optimal path solution process, thereby improving the placement rate of the in-line placement machine and making it more suitable for the application background. It has important theoretical and application value. Summary of the invention

[0013] The purpose of the present invention is to solve the problem of assembly efficiency caused by the existing in-line placement machine optimization technology for PCB placement, which has strong application scenario limitations, slow optimal path solution speed, and incomplete consideration of factors affecting assembly efficiency. The present invention provides a method for optimizing the surface mounting process of an in-line placement machine based on hierarchical heuristics. Among them, the assembly process consists of three processes: component type allocation process in the feeder slot, picking process and mounting process.

[0014] A method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics includes the following steps:

[0015] Step 1: Allocate each component type to be allocated and all components corresponding to the component type to a corresponding slot on the feeder base, specifically:

[0016] Step 1-1, set the initial value of M to 1, where M is the number of rounds of the scanning cycle;

[0017] Step 1-2, in the Mth round of scanning cycle, the slots on the feeder base are scanned from left to right in sequence by the in-line placement head group, and the optimal slot is allocated to the placement head at the head end of the in-line placement head group. After the placement head at the head end is aligned with the optimal slot, the corresponding component type and component quantity are allocated to the empty state slots corresponding to each placement head on the in-line placement head group; wherein the state of each slot on the feeder base is a non-empty state or an empty state;

[0018] Step 1-3, take the minimum value δ of the number of components in each slot corresponding to each placement head of the in-line placement head group, subtract δ from the number of components in each slot corresponding to each placement head of the in-line placement head group, and update the number of components in each slot; determine Whether there is ψ i '=0, the result is yes. At this time, all component types in I have been allocated to the corresponding slots on the feeder base, and step 2 is executed. The result is no, let M=M+1, and execute steps 1-2;

[0019] Where i′ is the component type index, I is the set of component type indexes, and ψ i ′ is the number of components of component type i′;

[0020] Step 2, in the Nth picking cycle, assigning a corresponding type of component to be picked up to each placement head according to the type of nozzle on each placement head of the in-line placement head group; wherein N is an integer, and the initial value of N is 1;

[0021] Step 3: Determine whether there is a component type to be picked up in the slot on the feeder base that has not yet been assigned a corresponding placement head. If the result is yes, set N=N+1 and execute step 2. Otherwise, execute step 4.

[0022] Step 4: According to the component type allocation result of each placement head on the in-line placement head group in each picking cycle, the optimal placement path under the current allocation result is obtained according to the nearest neighbor cluster search strategy, thereby completing the optimization of the placement process.

[0023] Preferably, step 1-2, scanning the slots on the feeder base from left to right in sequence by the in-line placement head group, and assigning the optimal slot to the placement head at the head end of the in-line placement head group in each scanning cycle includes:

[0024] Step 1-2-11, determining the component type pre-installed on each non-empty slot on the feeder base and the number of components corresponding to the component type pre-installed on each non-empty slot;

[0025] Step 1-2-12: Have the placement head at the head end of the in-line placement head group scan all slots on the feeder base from left to right in turn, and record the sum of the number of components in the non-empty slots corresponding to each placement head of the in-line placement head group when the placement head at the head end is aligned with each slot. At this time, the slot that is aligned with the placement head at the head end when the sum of the number of components in the non-empty slots is the largest is taken as the optimal slot.

[0026] Preferably, in step 1-2, after the placement head at the head end is aligned with the optimal slot, the corresponding component type and component quantity are allocated to the empty slots corresponding to each placement head on the in-line placement head group, and the implementation method includes the following steps:

[0027] Step 1-2-21, align the placement head at the head end with the optimal slot;

[0028] Step 1-2-22, determine whether there is a nozzle on the placement head corresponding to each empty slot relative to the in-line placement head group. If the result is yes, execute step 1-2-23; if the result is no, execute step 1-2-24;

[0029] Step 1-2-23, from left to right, traverse the nozzle types of the placement heads corresponding to all empty slots, and according to the nozzle type on the placement head corresponding to each empty slot, assign the corresponding component type and component quantity to the empty slot corresponding to the placement head, and end;

[0030] Step 1-2-24

[0031] Take argmax i′∈I {ψ i′ |∑ j′∈J j′·ξ i′j′ γ j′}, argmax i′∈I {ψ i′ |∑ j′∈J j′·ξ i′j′ γ j′} corresponds to the component type i′ and the number of components ψ i′ , assigned to the current empty state slot, end;

[0032] Among them, ξ i′j′ =1 means that the component type i′ matches the nozzle type j′, ξ i′j′ = 0 means that the component type i′ does not match the nozzle type j′; j′ is the nozzle type index, J is the set of nozzle type indexes, i′ is the component type index, I is the set of component type indexes, ψ i ′ is the number of components of component type i′, γ j ' represents the available number of nozzle type j'.

[0033] Preferably, in step 1-2-23, the nozzle types of the placement heads corresponding to all empty slots are traversed from left to right in sequence, and according to the nozzle type on the placement head corresponding to each empty slot, the corresponding component type and component quantity are allocated to the empty slot corresponding to the placement head according to the nozzle type, including the following steps:

[0034] Step 1-2-23-1, according to the nozzle type of the placement head corresponding to each empty slot, determine whether the number of component types that successfully match the nozzle type is greater than 0. If the result is yes, execute step 1-2-23-2; if the result is no, execute step 1-2-23-3;

[0035] Step 1-2-23-2, when the number of components corresponding to the component type that successfully matches the nozzle type corresponding to the current placement head is the largest, the corresponding component type and the number of all components corresponding to the component type are determined, and the corresponding component type and the number of all components corresponding to the component type are allocated to the empty state slot corresponding to the current placement head, and the process ends;

[0036] Step 1-2-23-3, for Take argmax i′∈I {ψ i′}, the corresponding component type i′ and the number of components ψ i′ Push into the stack S, count all unallocated empty state slots, pop each component type in the stack S from the stack S in turn, and allocate all the component quantities corresponding to each component type to a corresponding unallocated empty state slot, wherein each component type is popped out from the stack S in turn in the order of the number of components of each component type from large to small, and then end;

[0037] Among them, ξ i′j′ =1 means that the component type i′ matches the nozzle type j′, ξ i′j′ = 0 means that the component type i′ does not match the nozzle type j′; j′ is the nozzle type index, i′ is the component type index, I is the set of component type indexes, ψ i′ is the number of components of component type i′.

[0038] Preferably, the implementation method of step 2, in the Nth picking cycle, allocating the corresponding type of component to be picked up to each placement head according to the type of nozzle on each placement head of the in-line placement head group comprises the following steps:

[0039] Step 2-1, make the inline placement head group scan the slots on the feeder base from left to right in sequence. When the inline placement head group is in different scanning slots, calculate the initial value function value of each placement head on the inline placement head group; and judge whether the nozzle type on each placement head matches the component type in the slot aligned with the placement head. If the result is yes, the initial value function value of the placement head is used as the final value function value. If the result is no, a penalty value is assigned to the placement head, and the penalty value is superimposed on the initial value function value to be used as the final value function value of the placement head;

[0040] Step 2-2, the in-line placement head group is in different scanning slots, and the final value function value of each placement head is accumulated from left to right. When the accumulation is accumulated to the placement head h′, the accumulation result is greater than 0, and the component type is allocated to the placement head h′ corresponding to the current accumulation result, specifically: the component type in the slot aligned with the placement head h′ is allocated to the placement head h′; otherwise, when the accumulation is accumulated to the placement head h′, the accumulation result is less than or equal to 0, and the component type is not allocated to the placement head h′ corresponding to the current accumulation result; h′ is the placement head index;

[0041] Step 2-3, when the in-line placement head group is in different scanning slots, calculate the total benefit of the placement head allocation results of the in-line placement head group, and then select the maximum total benefit of the placement head allocation results from the entire scanning process of the in-line placement head group, and use the allocation result of the placement head of the allocated component type corresponding to the maximum total benefit of the placement head allocation result as the optimal allocation result;

[0042] Step 2-4, determine whether there is a placement head that has not been assigned a component type under the optimal allocation result of the in-line placement head group. If the result is yes, make each placement head that has not been assigned a component type on the in-line placement head group traverse the slots where the unassigned component types are located on the scan feeder base from left to right in turn, and select the maximum value of the initial value function value of each placement head that has not been assigned a component type on the in-line placement head group under different scanning slots, and assign the component type in the slot corresponding to the placement head when the initial value function value is the maximum to the current placement head, so as to achieve in the Nth picking cycle, all placement heads on the in-line placement head group are assigned the corresponding component type to be picked up.

[0043] Preferably, the initial value function value of each placement head is obtained by calculating the number of components in the slot aligned with the placement head using the value function.

[0044] Preferably, in step 2-3, the total revenue of the placement head allocation result of the in-line placement head group is: formed by the sum of the target revenue of the in-line placement head group and the long-term revenue of the in-line placement head group.

[0045] Preferably, the target revenue of the in-line placement head group is obtained as follows:

[0046] The target profit of the inline placement head group

[0047] in,

[0048]

[0049] σ is the number of nozzle changes resulting from the current component allocation result;

[0050] h′ is the placement head index, and H is the placement head index set;

[0051] j′ is the nozzle type index, and J is the set of nozzle type indexes;

[0052] N(h′) is the nozzle type on the placement head h′;

[0053] Match the component type on the placement head h′ with the nozzle type j′;

[0054] The component type on the placement head h′ does not match the nozzle type j′;

[0055] The type of component to be picked up has been assigned to the placement head h′;

[0056] The number of components corresponding to the component type assigned to be picked up on the placement head h′;

[0057] e 1 and e 2 are the first and second objective coefficients, respectively;

[0058] ω is the maximum number of components that can be picked up simultaneously by the in-line placement head group during the current picking process.

[0059] Preferably, the method for obtaining the long-term benefits of the in-line placement head group includes the following steps:

[0060] Step 2-3-21, let the long-term benefit of the in-line placement head group be calculated as V l The initial value is V l =-e 1 ·σ; where e 1 is the first target coefficient, σ is the number of nozzle changes resulting from the current component allocation result;

[0061] Step 2-3-22, order in,

[0062]

[0063] In the formula,

[0064] e 2 is the second target coefficient;

[0065] ω is the maximum number of components that can be picked up simultaneously by the in-line placement head group during the current picking process;

[0066] h′ is the placement head index, and H is the placement head index set;

[0067] is the number of components corresponding to the component type assigned to be picked up on the placement head h′;

[0068] Step 2-3-23, Update make

[0069] Step 2-3-24, determine for Does it exist? If the result is yes, end; if the result is no, proceed to step 2-3-22.

[0070] Preferably, in step 4, according to the component type allocation result of each placement head on the in-line placement head group in each picking cycle, the implementation method of obtaining the optimal placement path under the current allocation result according to the nearest neighbor cluster search strategy is:

[0071] Step 4-1, take the lower left corner of the mounting area of ​​the PCB board as the search starting point, and set the total number of mounting heads to b;

[0072] Step 4-2, traverse all the placement heads from left to right, select β placement points closest to the search starting point from the placement points corresponding to the placement head h′, and use the β placement points as the landmark points of the placement head h′; the number of components in the slots aligned with each placement head is the same as the number of placement points corresponding to the placement head; the initial value of h′ is 1, h′ is the placement head index; β is an integer;

[0073] Step 4-3, select the P placement points closest to each landmark point of the placement head h′ from the placement points corresponding to the placement head h′+1, and form P paths between each landmark point of the placement head h′ and the P closest placement points corresponding to it, so as to obtain β×P paths between the placement head 1 and the placement head h′+1, and select the β initial optimal paths with the shortest paths from the β×P paths as the β initial optimal paths between the placement head 1 and the placement head h′+1, and use the β placement points corresponding to the placement head h′+1 in the β initial optimal paths as the landmark points of the placement head h′+1; wherein, P≤β; P is an integer;

[0074] Step 4-4, determine whether the placement head h′+1 is equal to b, if yes, execute step 4-5, otherwise, set h′=h′+1, and execute step 4-3;

[0075] Step 4-5: Select a shortest path from the β initial optimal paths between the mounting head 1 and the mounting head b, and use the shortest path as the optimal path between the mounting head 1 and the mounting head b.

[0076] The beneficial effects brought by the present invention are:

[0077] The present invention proposes a method for optimizing the surface mounting process of a straight-row placement machine based on hierarchical heuristics, which uses the idea of ​​hierarchical heuristics to optimize the PCB assembly process. Hierarchical heuristics refers to dividing the problems existing in the entire assembly process into three sub-problems, and the solution of the previous sub-problem is used as the input of the solution of the next sub-problem, which reduces the search space of the problem; the hierarchical heuristics of the present invention includes three layers, namely, the first layer determines the component type assigned to the feeder slot, the second layer determines the component type assigned to the placement head, and the third layer determines the placement path. For specific logical ideas, see Figure 4 The present invention can better handle the optimization problem of PCB assembly process under complex circuit boards and different slots, obtain efficient surface mounting process solutions with less time and space costs, and meet the needs of actual production mounting tasks in various industrial environments. Specifically:

[0078] 1. The present invention takes into account the current state of each slot on the feeder base, and the state of each slot on the feeder base is a non-empty state or an empty state; that is: in the initial state, it takes into account the situation that some slots have corresponding component types, and the placement head corresponding to the slot where the component type is located; compared with the prior art, the present invention has strong applicability in application scenarios. When used specifically, there is no need to clear each slot on the feeder base and then reallocate components to each slot. Instead, based on the current state of each slot on the existing feeder base, component type allocation can be implemented for the empty slots to perform the next step. The present invention does not need to clear non-empty slots in the initial state and then reallocate components, which overcomes the problem of limitations in application scenarios in the prior art; the process of clearing all slots is omitted, the process of the optimization method is further simplified, and the assembly efficiency is further improved.

[0079] 2. The in-line placement machine surface mounting process optimization method based on hierarchical heuristics described in the present invention disassembles the surface mounting process optimization problem according to the progressive relationship. The hierarchical search process is essentially a process of "pruning" the problem. Pruning effectively improves the time and space efficiency of the search algorithm. The scanning method after pruning gradually compresses the search space during the execution process, reduces the amount of calculation in the search space, and thus improves the solution speed. The scanning process can optimize the picking efficiency of the surface mounting process, thereby improving the process of obtaining the optimal mounting path, achieving the improvement of the mounting efficiency, and ensuring that the boundary of the overall mounting efficiency search is maintained at a high level, thereby further improving the assembly efficiency; wherein, the specific "pruning" process can be reflected in the determination of the optimal slot allocation for the mounting head at the head end of the in-line mounting head group, the allocation of the component type of the empty state slot, the allocation of the corresponding component type to be picked up to each mounting head during the picking process, and the acquisition of the optimal mounting path. The solution of the previous sub-problem is used as the input of the solution of the next sub-problem, reducing the search space of the problem.

[0080] The present invention improves the search efficiency of surface mount process optimization problems. The hierarchical search framework proposed by the present invention structures the problem layer by layer, disassembles the problem from simple to complex in terms of complexity, and the construction heuristic optimization algorithm designed for each sub-problem also has a high solving efficiency. Compared with mathematical programming and evolutionary search algorithms, the method proposed by the present invention has a significant improvement in search efficiency.

[0081] 3. Since the type of components to be picked up assigned to the placement head during the picking process affects the overall performance, in order to maximize the overall assembly efficiency; Steps 2 and 3, in each picking cycle, according to the type of nozzles on each placement head on the in-line placement head group, assign the corresponding type of components to be picked up to each placement head to maximize the overall assembly efficiency and ultimately improve the placement efficiency.

[0082] 4. In the process of obtaining the optimal mounting path, the present invention comprehensively considers factors affecting the assembly efficiency, such as the number of simultaneous pick-ups of the mounting head group, the number of nozzle changes of the mounting head, and the length of the cantilever's moving path during the PCB assembly process of the in-line mounter, so that the optimal mounting path can be obtained quickly and accurately, thereby improving the mounting efficiency and ultimately achieving the purpose of improving the assembly efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 The figure is a schematic diagram of the structure of an in-line single cantilever placement machine; wherein reference numeral 1 represents a placement head, reference numeral 2 represents a fixed camera, reference numeral 3 represents an automatic nozzle changer, reference numeral 4 represents a feeder; reference numeral 5 represents a nozzle; X and Y are the movement directions of the cantilever on the X-axis and Y-axis respectively;

[0084] Figure 2It is the process flow chart of surface mounting process;

[0085] Figure 3 Optimize the problem relationship logic diagram for the surface mounting process;

[0086] Figure 4 It is a logic block diagram of the surface mounting process optimization method of the in-line placement machine based on hierarchical heuristics according to the present invention;

[0087] Figure 5 It is a flow chart of the method for optimizing the surface mounting process of a straight-row placement machine based on hierarchical heuristics according to the present invention;

[0088] Figure 6 Comparison chart of search efficiency for surface mount process optimization algorithms. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0091] Specific implementation method 1. Figure 5 The present embodiment is described. The surface mounting process optimization method of a straight-line placement machine based on hierarchical heuristics described in the present embodiment comprises the following steps:

[0092] Step 1: Allocate each component type to be allocated and all components corresponding to the component type to a corresponding slot on the feeder base, specifically:

[0093] Step 1-1, set the initial value of M to 1, where M is the number of rounds of the scanning cycle;

[0094] Step 1-2, in the Mth round of scanning cycle, the slots on the feeder base are scanned from left to right in sequence by the in-line placement head group, and the optimal slot is allocated to the placement head at the head end of the in-line placement head group. After the placement head at the head end is aligned with the optimal slot, the corresponding component type and component quantity are allocated to the empty state slots corresponding to each placement head on the in-line placement head group; wherein the state of each slot on the feeder base is a non-empty state or an empty state;

[0095] Step 1-3, take the minimum value δ of the number of components in each slot corresponding to each placement head of the in-line placement head group, subtract δ from the number of components in each slot corresponding to each placement head of the in-line placement head group, and update the number of components in each slot; determine Whether there is ψ i′ =0, the result is yes. At this time, all component types in I have been allocated to the corresponding slots on the feeder base, and step 2 is executed. The result is no, let M=M+1, and execute steps 1-2;

[0096] Where i′ is the component type index, I is the set of component type indexes, and ψ i′ is the number of components of component type i′;

[0097] Step 2, in the Nth picking cycle, assigning a corresponding type of component to be picked up to each placement head according to the type of nozzle on each placement head of the in-line placement head group; wherein N is an integer, and the initial value of N is 1;

[0098] Step 3: Determine whether there is a component type to be picked up in the slot on the feeder base that has not yet been assigned a corresponding placement head. If the result is yes, set N=N+1 and execute step 2. Otherwise, execute step 4.

[0099] Step 4: According to the component type allocation result of each placement head on the in-line placement head group in each picking cycle, the optimal placement path under the current allocation result is obtained according to the nearest neighbor cluster search strategy, thereby completing the optimization of the placement process.

[0100] In specific applications, each slot on the feeder base is used to store a component type and the total number of components corresponding to the component type. In the present invention, i' is the component type index, which is a variable. For example, the index value can be 1, 2, 3, 4... etc., but is not limited to this value method.

[0101] The concept of the present invention is as follows: first, each component type is assigned to a corresponding slot, and the status information of each slot is considered in the process of assigning the corresponding slot, and the slot is assigned under the constraint condition of assigning the optimal slot to the placement head located at the head end of the in-line placement head group; secondly, the corresponding component type is assigned to each placement head, and finally, according to the nearest neighbor cluster search strategy, according to the constraint condition of the component type assignment result of each placement head on the in-line placement head group in each picking cycle, the optimal placement path under the current assignment result is obtained.

[0102] In step 1, the situation in which some slots have been pre-installed with corresponding component types and the placement heads corresponding to the slots where the component types are located are taken into consideration in the initial state; compared with the prior art, the present invention has strong applicability in application scenarios. When used specifically, there is no need to clear the slots on the feeder base and then reallocate components to the slots. Instead, based on the current state of the slots on the existing feeder base, the component types can be subsequently allocated to the empty slots to carry out the next step. The present invention does not need to clear the non-empty slots in the initial state and then reallocate components, which overcomes the problem of limitations in application scenarios in the prior art.

[0103] Since the type of components to be picked up assigned to the placement head during the picking process affects the overall performance, in order to maximize the overall assembly efficiency; Step 2 and Step 3, in each picking cycle, according to the type of nozzle on each placement head on the in-line placement head group, assign the corresponding type of components to be picked up to each placement head to maximize the overall assembly efficiency and ultimately improve the placement efficiency.

[0104] It is further provided that, in step 1-2, the slots on the feeder base are scanned from left to right in sequence by the in-line placement head group, and the optimal slot is allocated to the placement head at the head end of the in-line placement head group in each scanning cycle, including:

[0105] Step 1-2-11, determining the component type pre-installed on each non-empty slot on the feeder base and the number of components corresponding to the component type pre-installed on each non-empty slot;

[0106] Step 1-2-12: Have the placement head at the head end of the in-line placement head group scan all slots on the feeder base from left to right in turn, and record the sum of the number of components in the non-empty slots corresponding to each placement head of the in-line placement head group when the placement head at the head end is aligned with each slot. At this time, the slot that is aligned with the placement head at the head end when the sum of the number of components in the non-empty slots is the largest is taken as the optimal slot.

[0107] In the present preferred implementation, a specific implementation method is provided for allocating the optimal slot to the placement head located at the head end of the in-line placement head group in each scanning cycle. The specific process takes into account the state of the head group alignment slot, that is, the type and number of components assigned to its slot, the number of empty slots to be assigned, the consistency of the nozzle types of the head group components and the assigned components, etc., so that the optimal slot selected by the present preferred implementation scheme meets the requirement of maximizing the number of placement points allocated by the placement head group under the nozzle type constraint.

[0108] Furthermore, it is provided that in step 1-2, after the placement head at the head end is aligned with the optimal slot, the corresponding component type and component quantity are allocated to the empty slot corresponding to each placement head on the in-line placement head group, and the implementation method includes the following steps:

[0109] Step 1-2-21, align the placement head at the head end with the optimal slot;

[0110] Step 1-2-22, determine whether there is a nozzle on the placement head corresponding to each empty slot relative to the in-line placement head group. If the result is yes, execute step 1-2-23; if the result is no, execute step 1-2-24;

[0111] Step 1-2-23, from left to right, traverse the nozzle types of the placement heads corresponding to all empty slots, and according to the nozzle type on the placement head corresponding to each empty slot, assign the corresponding component type and component quantity to the empty slot corresponding to the placement head, and end;

[0112] Step 1-2-24

[0113] Take argmax i′∈I {ψ i′ |∑ j′∈J j′·ξ i′j′ γ j′}, argmax i′∈I {ψ i′ |∑ j′∈J j′·ξ i′j′ γ j′} corresponds to the component type i′ and the number of components ψ i′ , assigned to the current empty state slot, end;

[0114] Among them, ξ i′j′ =1 means that the component type i′ matches the nozzle type j′, ξ i′j′ = 0 means that the component type i′ does not match the nozzle type j′; j′ is the nozzle type index, J is the set of nozzle type indexes, i′ is the component type index, I is the set of component type indexes, ψ i′ is the number of components of component type i′, γ j′ Indicates the available number of nozzle type j'.

[0115] In this preferred embodiment, components are allocated to empty slots of each placement head, taking into account the type constraints of components that have been allocated to the placement head group alignment slots, ensuring that the slots with components of similar numbers are allocated relatively concentrated.

[0116] Furthermore, in step 1-2-23, the nozzle types of the placement heads corresponding to all empty slots are traversed from left to right in turn, and according to the nozzle type on the placement head corresponding to each empty slot, the corresponding component type and component quantity are allocated to the empty slot corresponding to the placement head, including the following steps:

[0117] Step 1-2-23-1, according to the nozzle type of the placement head corresponding to each empty slot, determine whether the number of component types that successfully match the nozzle type is greater than 0. If the result is yes, execute step 1-2-23-2; if the result is no, execute step 1-2-23-3;

[0118] Step 1-2-23-2, when the number of components corresponding to the component type that successfully matches the nozzle type corresponding to the current placement head is the largest, the corresponding component type and the number of all components corresponding to the component type are determined, and the corresponding component type and the number of all components corresponding to the component type are allocated to the empty state slot corresponding to the current placement head, and the process ends;

[0119] Step 1-2-23-3, for Take argmax i′∈I {ψ i′}, the corresponding component type i′ and the number of components ψ i′ Push onto stack And count all unassigned empty state slots, stack Each component type in the stack is Pop out and assign all the component quantities corresponding to each component type to a corresponding unassigned empty state slot, where each component type is popped out of the stack in turn. The order in which the components are popped out is from large to small according to the number of components of each component type, and the process ends;

[0120] Among them, ξ i′j′ =1 means that the component type i′ matches the nozzle type j′, ξ i′j′ = 0 means that the component type i′ does not match the nozzle type j′; j′ is the nozzle type index, i′ is the component type index, I is the set of component type indexes, ψ i′ is the number of components of component type i′.

[0121] In the preferred embodiment, in the process of allocating components to the corresponding empty slots of each placement head on the in-line placement head group, the allocation information of the component types pre-installed in certain slots on the base can be utilized to the greatest extent, so that the positions of the component types allocated to the slots are relatively concentrated, providing a reference for the allocation of components on the placement head mechanism.

[0122] Furthermore, it is given that, in step 2, in the Nth picking cycle, according to the nozzle type on each placement head of the in-line placement head group, the implementation method of allocating the corresponding type of component to be picked up to each placement head includes the following steps:

[0123] Step 2-1, make the inline placement head group scan the slots on the feeder base from left to right in sequence. When the inline placement head group is in different scanning slots, calculate the initial value function value of each placement head on the inline placement head group; and judge whether the nozzle type on each placement head matches the component type in the slot aligned with the placement head. If the result is yes, the initial value function value of the placement head is used as the final value function value. If the result is no, a penalty value is assigned to the placement head, and the penalty value is superimposed on the initial value function value to be used as the final value function value of the placement head;

[0124] Step 2-2, the in-line placement head group is in different scanning slots, and the final value function value of each placement head is accumulated from left to right. When the accumulation is accumulated to the placement head h′, the accumulation result is greater than 0, and the component type is allocated to the placement head h′ corresponding to the current accumulation result, specifically: the component type in the slot aligned with the placement head h′ is allocated to the placement head h′; otherwise, when the accumulation is accumulated to the placement head h′, the accumulation result is less than or equal to 0, and the component type is not allocated to the placement head h′ corresponding to the current accumulation result; h′ is the placement head index;

[0125] Step 2-3, when the in-line placement head group is in different scanning slots, calculate the total benefit of the placement head allocation results of the in-line placement head group, and then select the maximum total benefit of the placement head allocation results from the entire scanning process of the in-line placement head group, and use the allocation result of the placement head of the allocated component type corresponding to the maximum total benefit of the placement head allocation result as the optimal allocation result;

[0126] Step 2-4, determine whether there is a placement head that has not been assigned a component type under the optimal allocation result of the in-line placement head group. If the result is yes, make each placement head that has not been assigned a component type on the in-line placement head group traverse the slots where the unassigned component types are located on the scan feeder base from left to right in turn, and select the maximum value of the initial value function value of each placement head that has not been assigned a component type on the in-line placement head group under different scanning slots, and assign the component type in the slot corresponding to the placement head when the initial value function value is the maximum to the current placement head, so as to achieve in the Nth picking cycle, all placement heads on the in-line placement head group are assigned the corresponding component type to be picked up.

[0127] In this preferred embodiment, allocating components to be picked up to the placement heads can ensure that when the placement head group picks up components from the feeder slots, more simultaneous picks can be obtained at a smaller nozzle replacement cost, thereby reducing the movement path of the placement head group during the picking process and improving the picking efficiency.

[0128] In step 2-1, when the in-line head group scans the slots on the feeder base from left to right in sequence, it is equivalent to considering the "in-line head group" as a whole. In specific applications, for example: when the placement head at the head end of the in-line head group is aligned with the first slot, it is regarded as the in-line head group scanning the first slot. The slots on the feeder base are sorted from left to right in sequence.

[0129] Furthermore, the initial value function value of each placement head is obtained by calculating the number of components in the slot aligned with the placement head using the value function. The larger the number of components in the slot aligned with the placement head, the larger the corresponding initial value function value. The value function is a conventional technical means in the prior art.

[0130] Furthermore, in step 2-3, the total benefit of the placement head allocation result of the in-line placement head group is: composed of the sum of the target benefit of the in-line placement head group and the long-term benefit of the in-line placement head group.

[0131] Furthermore, a specific implementation method of the target benefit of the in-line placement head group is given. In specific application, the target benefit of the in-line placement head group is obtained as follows:

[0132] The target profit of the inline placement head group

[0133] in,

[0134]

[0135] σ is the nozzle replacement rate resulting from the current component allocation result;

[0136] h′ is the placement head index, and H is the placement head index set;

[0137] j′ is the nozzle type index, and J is the set of nozzle type indexes;

[0138] N(h′) is the nozzle type on the placement head h′;

[0139] Match the component type on the placement head h′ with the nozzle type j′;

[0140] The component type on the placement head h′ does not match the nozzle type j′;

[0141] The type of component to be picked up has been assigned to the placement head h′;

[0142] is the number of components corresponding to the component type assigned to be picked up on the placement head h′;

[0143] e 1 and e 2 are the first and second objective coefficients, respectively;

[0144] ω is the maximum number of components that can be picked up simultaneously by the in-line placement head group during the current picking process.

[0145] In this preferred implementation scheme, the impact of factors such as the number of simultaneous picks and nozzle replacement on the target benefit is taken into account, and a forward-looking search that combines long-term and short-term benefits ensures that the overall solution will not fall into a local optimum.

[0146] Furthermore, it is given that the long-term benefits of the in-line placement head group can be obtained by the following steps:

[0147] Step 2-3-21, let the long-term benefit of the in-line placement head group be calculated as V l The initial value is V l =-e 1 ·σ; where e 1 is the first target coefficient, σ is the nozzle replacement rate generated by the current component allocation result;

[0148] Step 2-3-22, order in,

[0149]

[0150] In the formula,

[0151] e 2 is the second target coefficient;

[0152] ω is the maximum number of components that can be picked up simultaneously by the in-line placement head group during the current picking process;

[0153] h′ is the placement head index, and H is the placement head index set;

[0154] is the number of components corresponding to the component type assigned to be picked up on the placement head h′;

[0155] Step 2-3-23, Update make

[0156] Step 2-3-24, determine for Does it exist? If the result is yes, end; if the result is no, proceed to step 2-3-22.

[0157] In the preferred embodiment, the goal of calculating the long-term benefits is the improvement in picking efficiency brought about by each placement head picking up all placement points from the alignment slots, which also refers to the reduction in the number of pick-ups by the placement head group; and in the process of obtaining long-term benefits, the impact of the picking results themselves on the efficiency, the impact of the current picking results on the subsequent picking process, the impact of the current picking result nozzle replacement on the overall picking efficiency, etc. are taken into account, thereby improving the efficiency of the entire picking process.

[0158] Furthermore, it is given that in step 4, according to the component type allocation result of each placement head on the in-line placement head group in each picking cycle, the implementation method of obtaining the optimal placement path under the current allocation result according to the nearest neighbor cluster search strategy is:

[0159] Step 4-1, take the lower left corner of the mounting area of ​​the PCB board as the search starting point, and set the total number of mounting heads to b;

[0160] Step 4-2, traverse all the placement heads from left to right, select β placement points closest to the search starting point from the placement points corresponding to the placement head h′, and use the β placement points as the landmark points of the placement head h′; the number of components in the slots aligned with each placement head is the same as the number of placement points corresponding to the placement head; the initial value of h′ is 1, h′ is the placement head index; β is an integer;

[0161] Step 4-3, select the P placement points closest to each landmark point of the placement head h′ from the placement points corresponding to the placement head h′+1, and form P paths between each landmark point of the placement head h′ and the P closest placement points corresponding to it, so as to obtain β×P paths between the placement head 1 and the placement head h′+1, and select the β initial optimal paths with the shortest paths from the β×P paths as the β initial optimal paths between the placement head 1 and the placement head h′+1, and use the β placement points corresponding to the placement head h′+1 in the β initial optimal paths as the landmark points of the placement head h′+1; wherein, P≤β; P is an integer;

[0162] Step 4-4, determine whether the placement head h′+1 is equal to b, if yes, execute step 4-5, otherwise, set h′=h′+1, and execute step 4-3;

[0163] Step 4-5: Select a shortest path from the β initial optimal paths between the mounting head 1 and the mounting head b, and use the shortest path as the optimal path between the mounting head 1 and the mounting head b.

[0164] In this preferred embodiment, according to the existing component allocation results, the "single direction" search strategy is used to determine the placement points processed by the placement heads in each pick-up cycle, and the placement order within the cycle is optimized by dynamic programming, so as to achieve the optimization of the placement path. Through the β initial optimal paths between placement head 1 and placement head h′+1, the β initial optimal paths between placement head 1 and placement head b are obtained by continuously "pruning" through a hierarchical search method (that is, "pruning" the problem), and the shortest path among the β initial optimal paths is selected as the optimal path corresponding to all placement heads. The pruned scanning method gradually compresses the search space during execution, reduces the amount of calculation in the search space, and thus improves the solution speed.

[0165] Verification test:

[0166] The following examples are used to verify the beneficial effects of the present invention, specifically:

[0167] This embodiment will use the data in PCB1 to PCB10 to verify the optimization effect and search efficiency of the in-line placement machine surface mounting process optimization method based on hierarchical heuristics proposed in the present invention. The embodiment of the present invention is the commonly used production data on the PCB production line, in which the data with a large number of mounting points can also represent the situation of multiple small batch orders under the fixed feeder (i.e. fixed slot) configuration. Table 1 shows the relevant parameters of the PCB data.

[0168] Table 1 PCB comparison data parameters

[0169]

[0170] In order to further illustrate the superiority of the surface mounting process optimization method of the in-line placement machine based on hierarchical heuristic proposed in the present invention, the optimization method proposed in the present invention is compared horizontally with the current mainstream aggregate mathematical programming method, hybrid genetic algorithm and industrial optimization solver which are also heuristic. The comparison is mainly carried out around the main performance indicators such as the number of picking and pasting cycles, the number of nozzle changes and the number of simultaneous picks. The comparison results are shown in Table 2.

[0171] Table 2 Comparison of main performance indicators of the method proposed in this invention with industrial solvers, aggregate mathematical programming and hybrid genetic algorithm

[0172]

[0173] Table 2 corresponds to the number of pick-and-place cycles, number of nozzle changes and number of picks for different methods. The number of pick-and-place cycles is one of the important sub-goals that reflects the overall assembly efficiency, because it will affect the length of the cantilever's moving route. A large number of pick-up cycles will increase the number of round trips between the cantilever and the feeder base. The optimization method proposed in the present invention shows a stronger search capability when processing complex data. The aggregate mathematical programming method completely ignores the simultaneous material collection factor, so the overall performance index is poor, especially in the indicator of the number of picks, which is obviously inferior to other methods. It cannot guarantee the solution of verification data of different scales, so its main performance indicators are not compared with other methods.

[0174] The present invention proposes considerations for simultaneous picking. The hybrid genetic algorithm constructs a simultaneous picking group to improve the picking efficiency, and the overall average number of picking times is significantly better than other methods. However, it assumes that there is no nozzle replacement action in the surface mounting process, which directly leads to the cantilever frequently going back and forth between the PCB and the feeder base, resulting in redundant travel, which can be seen from the number of picking cycles.

[0175] Table 3 Comparison of assembly efficiency of the proposed method with industrial solvers, aggregate mathematical programming and hybrid genetic algorithm

[0176]

[0177] The present invention further compares the actual mounting efficiency of different methods and verifies the algorithm on the self-developed placement machine platform. Each method is applied three times on a batch of PCBs. The average assembly time results are shown in Table 3. 1 、E 2 、E 3 、E 4 They represent the assembly efficiency of the proposed method, industrial solver, aggregate mathematical programming and hybrid genetic algorithm, ΔE i =(E 1 -E i ) / E 1 ×100% (i=2, 3, 4), it means that the method proposed in the present invention brings about performance improvement compared with the corresponding method. As can be seen from Table 3, the method proposed in the present invention improves the assembly efficiency by 9.81%, 36.08% and 13.42% respectively compared with the industrial solver, aggregate mathematical programming and hybrid genetic algorithm. Although the method proposed in the present invention does not significantly exceed the results of the industrial solver on small and medium-sized data, as the scale of the problem increases, the advantages of the method proposed in the present invention become more obvious. Overall, the output results of the method proposed in the present invention are more stable, and the aggregate mathematical programming method and the hybrid genetic algorithm are significantly optimized, which further verifies that the method of the present invention using the heuristic search idea can bring more obvious performance improvement compared with the evolutionary algorithm and the mathematical programming method.

[0178] See also Figure 6 The present invention uses Python 3.8 to implement different optimization methods on a desktop computer equipped with an Intel Core i7 1.8-GHz CPU to compare the search efficiency of different optimization methods, and sets the search time limit to 10 minutes. Since the implementation method of the built-in industrial solver is not disclosed, this application only compares the method of the present invention, the aggregate mathematical programming method and the hybrid genetic algorithm. The average search time of the method proposed in the present invention is 24.21 seconds, while the average search time of the mathematical programming method is 97.767 seconds, and the average search time of the evolutionary algorithm is 944.195 seconds. For large-scale PCBs, the evolutionary algorithm takes a long time to find a solution, and the results of its random exploration are usually unstable. Although the aggregate mathematical programming method uses efficient aggregation-based technology to improve the search efficiency, it still cannot solve large-scale PCB data, which is an inherent defect of the mathematical programming method itself.

[0179] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. A method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics, characterized in that: The method comprises the following steps: Step 1: Allocate each component type to be allocated and all components corresponding to the component type to a corresponding slot on the feeder base, specifically: Step 1-1, set the initial value of M to 1, where M is the number of rounds of the scanning cycle; Step 1-2, in the Mth round of scanning cycle, the slots on the feeder base are scanned from left to right in sequence by the in-line placement head group, and the optimal slot is allocated to the placement head at the head end of the in-line placement head group. After the placement head at the head end is aligned with the optimal slot, the corresponding component type and component quantity are allocated to the empty state slots corresponding to each placement head on the in-line placement head group; wherein the state of each slot on the feeder base is a non-empty state or an empty state; Step 1-3, take the minimum value δ of the number of components in each slot corresponding to each placement head of the in-line placement head group, subtract δ from the number of components in each slot corresponding to each placement head of the in-line placement head group, and update the number of components in each slot; determine Whether there is ψ i′ =0, the result is yes. At this time, all component types in I have been allocated to the corresponding slots on the feeder base, and step 2 is executed. The result is no, let M=M+1, and execute steps 1-2; Where i′ is the component type index, I is the set of component type indexes, and ψ i′ is the number of components of component type i′; Step 2, in the Nth picking cycle, assigning a corresponding type of component to be picked up to each placement head according to the type of nozzle on each placement head of the in-line placement head group; wherein N is an integer, and the initial value of N is 1; Step 3: Determine whether there is a component type to be picked up in the slot on the feeder base that has not yet been assigned a corresponding placement head. If the result is yes, set N=N+1 and execute step 2. Otherwise, execute step 4. Step 4: According to the component type allocation result of each placement head on the in-line placement head group in each picking cycle, the optimal placement path under the current allocation result is obtained according to the nearest neighbor cluster search strategy, thereby completing the optimization of the placement process; The implementation method of obtaining the optimal placement path under the current allocation result is: Step 4-1, take the lower left corner of the mounting area of ​​the PCB board as the search starting point, and set the total number of mounting heads to b; Step 4-2, traverse all the placement heads from left to right, select β placement points closest to the search starting point from the placement points corresponding to the placement head h′, and use the β placement points as the landmark points of the placement head h′; the number of components in the slots aligned with each placement head is the same as the number of placement points corresponding to the placement head; the initial value of h′ is 1, h′ is the placement head index; β is an integer; Step 4-3, select the P placement points closest to each landmark point of the placement head h′ from the placement points corresponding to the placement head h′+1, and form P paths between each landmark point of the placement head h′ and the P closest placement points corresponding to it, so as to obtain β×P paths between the placement head 1 and the placement head h′+1, and select the β initial optimal paths with the shortest paths from the β×P paths as the β initial optimal paths between the placement head 1 and the placement head h′+1, and use the β placement points corresponding to the placement head h′+1 in the β initial optimal paths as the landmark points of the placement head h′+1; wherein, P≤β; P is an integer; Step 4-4, determine whether the placement head h′+1 is equal to b, if yes, execute step 4-5, otherwise, set h′=h′+1, and execute step 4-3; Step 4-5: Select a shortest path from the β initial optimal paths between the mounting head 1 and the mounting head b, and use the shortest path as the optimal path between the mounting head 1 and the mounting head b.

2. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 1 is characterized in that: Step 1-2, scanning the slots on the feeder base from left to right in sequence by the inline placement head group, and assigning the optimal slot to the placement head at the head end of the inline placement head group in each scanning cycle includes: Step 1-2-11, determining the component type pre-installed on each non-empty slot on the feeder base and the number of components corresponding to the component type pre-installed on each non-empty slot; Step 1-2-12: Have the placement head at the head end of the in-line placement head group scan all slots on the feeder base from left to right in turn, and record the sum of the number of components in the non-empty slots corresponding to each placement head of the in-line placement head group when the placement head at the head end is aligned with each slot. At this time, the slot that is aligned with the placement head at the head end when the sum of the number of components in the non-empty slots is the largest is taken as the optimal slot.

3. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 1 is characterized in that: In step 1-2, after the placement head at the head end is aligned with the optimal slot, the corresponding component type and component quantity are allocated to the empty slots corresponding to each placement head on the in-line placement head group. The implementation method includes the following steps: Step 1-2-21, align the placement head at the head end with the optimal slot; Step 1-2-22, determine whether there is a nozzle on the placement head corresponding to each empty slot relative to the in-line placement head group. If the result is yes, execute step 1-2-23; if the result is no, execute step 1-2-24; Step 1-2-23, from left to right, traverse the nozzle types of the placement heads corresponding to all empty slots, and according to the nozzle type on the placement head corresponding to each empty slot, assign the corresponding component type and component quantity to the empty slot corresponding to the placement head, and end; Step 1-2-24 Take argmax i′∈I {ψ i′ |∑ j′∈J j′·ξ i′j′ γ j′ }, will take argmax i′∈I {ψ i′ |∑ j′∈J j′·ξ i′j′ γ j′ } corresponds to the component type i′ and the number of components ψ i′ , assigned to the current empty state slot, end; Among them, ξ i′j′ =1 means that the component type i′ matches the nozzle type j′, ξ i′j′ = 0 means that the component type i′ does not match the nozzle type j′; j′ is the nozzle type index, J is the set of nozzle type indexes, i′ is the component type index, I is the set of component type indexes, ψ i′ is the number of components of component type i′, γ j′ Indicates the available number of nozzle type j'.

4. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 3 is characterized in that: In step 1-2-23, the nozzle types of the placement heads corresponding to all empty slots are traversed from left to right in turn, and according to the nozzle type on the placement head corresponding to each empty slot, the corresponding component type and component quantity are allocated to the empty slot corresponding to the placement head, including the following steps: Step 1-2-23-1, according to the nozzle type of the placement head corresponding to each empty slot, determine whether the number of component types that successfully match the nozzle type is greater than 0. If the result is yes, execute step 1-2-23-2; if the result is no, execute step 1-2-23-3; Step 1-2-23-2, when the number of components corresponding to the component type that successfully matches the nozzle type corresponding to the current placement head is the largest, the corresponding component type and the number of all components corresponding to the component type are determined, and the corresponding component type and the number of all components corresponding to the component type are allocated to the empty state slot corresponding to the current placement head, and the process ends; Step 1-2-23-3, for Take argmax i′∈I {ψ i′ }, the corresponding component type i′ and the number of components ψ i′ Push onto stack And count all unassigned empty state slots, stack Each component type in the stack is Pop out and assign all the component quantities corresponding to each component type to a corresponding unassigned empty state slot, where each component type is popped out of the stack in turn. The order in which the components are popped out is from large to small according to the number of components of each component type, and the process ends; Among them, ξ i′j′ =1 means that the component type i′ matches the nozzle type j′, ξ i′j′ = 0 means that the component type i′ does not match the nozzle type j′; j′ is the nozzle type index, i′ is the component type index, I is the set of component type indexes, ψ i′ is the number of components of component type i′.

5. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 1 is characterized in that: Step 2: In the Nth picking cycle, according to the nozzle type of each placement head on the in-line placement head group, the implementation method of allocating the corresponding type of component to be picked up to each placement head includes the following steps: Step 2-1, make the inline placement head group scan the slots on the feeder base from left to right in sequence. When the inline placement head group is in different scanning slots, calculate the initial value function value of each placement head on the inline placement head group; and judge whether the nozzle type on each placement head matches the component type in the alignment slot of the placement head. If the result is yes, the initial value function value of the placement head is used as the final value function value. If the result is no, a penalty value is assigned to the placement head, and the penalty value is superimposed on the initial value function value to be used as the final value function value of the placement head; Step 2-2, the in-line placement head group is in different scanning slots, and the final value function value of each placement head is accumulated from left to right. When the accumulation is accumulated to the placement head h′, if the accumulation result is greater than 0, the component type is allocated to the placement head h′ corresponding to the current accumulation result, specifically: the component type in the slot aligned with the placement head h′ is allocated to the placement head h′; otherwise, when the accumulation is accumulated to the placement head h′, if the accumulation result is less than or equal to 0, the component type is not allocated to the placement head h′ corresponding to the current accumulation result; h′ is the placement head index; Step 2-3, when the in-line placement head group is in different scanning slots, calculate the total benefit of the placement head allocation results of the in-line placement head group, and then select the maximum total benefit of the placement head allocation results from the entire scanning process of the in-line placement head group, and use the allocation result of the placement head of the allocated component type corresponding to the maximum total benefit of the placement head allocation result as the optimal allocation result; Step 2-4, determine whether there is a placement head that has not been assigned a component type under the optimal allocation result of the in-line placement head group. If the result is yes, make each placement head that has not been assigned a component type on the in-line placement head group traverse the slots where the unassigned component types are located on the scan feeder base from left to right in turn, and select the maximum value of the initial value function value of each placement head that has not been assigned a component type on the in-line placement head group under different scanning slots, and assign the component type in the slot corresponding to the placement head when the initial value function value is the maximum to the current placement head, so as to achieve in the Nth picking cycle, all placement heads on the in-line placement head group are assigned the corresponding component type to be picked up.

6. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 5 is characterized in that: The initial value of the value of each placement head is obtained by calculating the number of components in the slot aligned with the placement head using the value function.

7. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 5 is characterized in that: In step 2-3, the total benefit of the placement head allocation result of the in-line placement head group is: the sum of the target benefit of the in-line placement head group and the long-term benefit of the in-line placement head group.

8. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 7 is characterized in that: The target revenue of the in-line placement head group is obtained as follows: The target profit of the inline placement head group in, σ is the number of nozzle changes resulting from the current component allocation result; h′ is the placement head index, and H is the placement head index set; j′ is the nozzle type index, and J is the set of nozzle type indexes; N(h′) is the nozzle type on the placement head h′; Match the component type on the placement head h′ with the nozzle type j′; The component type on the placement head h′ does not match the nozzle type j′; The type of component to be picked up has been assigned to the placement head h′; is the number of components corresponding to the component type assigned to be picked up on the placement head h′; e1 and e2 are the first and second objective coefficients respectively; ω is the maximum number of components that can be picked up simultaneously by the in-line placement head group during the current picking process.

9. The method for optimizing the surface mounting process of a straight-line placement machine based on hierarchical heuristics according to claim 7, characterized in that: The long-term benefits of the in-line placement head group include the following steps: Step 2-3-21, let the long-term benefit of the in-line placement head group V l The initial value of is -e1·σ; where e1 is the first target coefficient, and σ is the number of nozzle changes resulting from the current component allocation result; Step 2-3-22, order in, In the formula, e2 is the second target coefficient; ω is the maximum number of components that can be picked up simultaneously by the in-line placement head group during the current picking process; h′ is the placement head index, and H is the placement head index set; is the number of components corresponding to the component type assigned to be picked up on the placement head h′; Step 2-3-23, Update make Step 2-3-24, determine for Does it exist? If the result is yes, end; if the result is no, proceed to step 2-3-22.

Images